----- Original Message -----
From: "naveeda salam" <[email protected]>
To: <[email protected]>
Sent: Thursday, April 24, 2003 4:14 AM
Subject: st: heteroskedasticity,autocorrelation and fixed effects
> Hi,
> i am trying to calculate the productivity at the
> industry and firm level data. According to the
> literature, productivity is taken as the unexplained
> part ie. residual.
>
> Since i am working with fixed effects model and taking
> the u, the fixed effects as indicators of
> productivity, of a specific industry or firm.
>
> Working with Xtregar-fe command, i see that rho-ar has
> returned with a very high value ie. .71, i suppose
> this is indicative of autocorrelation in residulas,
> and high number means high degree of autocorrelation.
> My question is , does the output with Xtregar-fe, has
> corrected for this autocorrelation, and if not , how
> to do it?
Yes,
After rho is estimated it uses a Cochrane-Orcutt method on each panel. You
notice that the number of obersvations has fallen by the number of panels.
> Secondly, while working with xtregar, is there any
> test which can predict for heteroskedasticity and any
> method which can correct for it? i think, lrtest is
> possible after xtgls but not with xtregar?
> I am interested in extracting fixed effects but at the
> same time correcting for autocorrelation and
> heteroskedasticity!
There is a potential identification problem using the fixed effects as a
measure of efficiency. Any time-invariant firm specific effects will show
up in the fixed effects estimates, confounding the efficiency estimates.
You might want to take a look at -xtfrontier-
Scott
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
